The history of the SOD1 mouse model begins in the early 1990s, shortly after mutations in the SOD1 gene were discovered to cause a familial form of ALS. At the time, this was a major breakthrough. ALS had long appeared mysterious and biologically inaccessible, but now there was at least one clearly identified genetic cause.
Researchers quickly engineered mice carrying mutant human SOD1 genes. The most famous became the G93A mouse, which overexpressed a mutant SOD1 protein and developed progressive paralysis followed by death. For the first time, scientists had an animal that visibly resembled ALS in a laboratory setting.
This transformed the field almost overnight.
Before that, ALS research had suffered from a lack of experimental models. Human motor neurons are difficult to study directly, and there was no easy way to reproduce the disease in animals. The SOD1 mouse solved that problem. Suddenly researchers could observe degeneration, test drugs, measure survival curves, and publish quantitative results.
The model became the backbone of preclinical ALS research.
But from the beginning there were warning signs.
Only a small fraction of ALS patients carry SOD1 mutations. Even in familial ALS, SOD1 represents only one subgroup. Most ALS patients instead show pathology dominated by TARDBP abnormalities, especially TDP-43 aggregation and nuclear depletion. Those findings became increasingly clear during the 2000s, but by then the SOD1 mouse infrastructure was already deeply entrenched.
Entire laboratories had been built around it. Grant systems expected it. Pharmaceutical companies used it as the standard entry gate for therapies. Reviewers trusted it because everyone else did.
And initially, it seemed to work.
Many interventions extended survival in SOD1 mice. Antioxidants helped. Anti-inflammatory approaches helped. Mitochondrial stabilizers helped. Gene silencing approaches helped. Dozens upon dozens of compounds produced promising mouse data.
Then the human trials began failing.
Again and again.
Some therapies showed no benefit whatsoever. Others produced tiny statistical effects with no meaningful clinical improvement. Many failed despite apparently robust preclinical evidence. The field responded by refining protocols, improving statistics, standardizing breeding methods, and increasing rigor in animal studies.
But the deeper problem was more uncomfortable to acknowledge.
What if the model itself was wrong for most ALS?
Over time, it became increasingly obvious that mutant SOD1 disease behaves differently from the more common TDP-43 forms of ALS. Even pathologically, they differ. Classical TDP-43 inclusions seen in sporadic ALS are often absent in SOD1 patients. The molecular cascades are not identical. The cellular stress responses differ. The protein aggregation behavior differs.
In effect, the field had spent decades optimizing therapies for one relatively rare subtype while hoping the results would generalize to all ALS cases.
The SOD1 mouse still had value. It was not useless. It taught researchers much about neuroinflammation, glial involvement, axonal degeneration, and protein misfolding toxicity. More recently, antisense therapies directly targeting mutant SOD1 finally produced some success precisely because they were aimed at the correct patient population rather than ALS as a whole.
But the broader lesson remains important.
A disease model is not the disease itself.
Once a scientific field becomes overly dependent on a single model organism, institutional momentum distorts reality. Funding flows toward what is measurable. Careers form around familiar techniques. Researchers learn to succeed within the model rather than question whether it truly represents the human condition.
And so ALS research spent decades curing mice while human patients continued to die.
***
The drug development pipeline in ALS has a structural flaw that borders on absurdity. Potential treatments are routinely screened in transgenic SOD1 mouse models and ranked based on whether they prolong survival in those animals. The problem is that SOD1 ALS represents only a small minority of human ALS cases, and even there, the disease mechanism is highly artificial compared to sporadic TDP-43 ALS. Yet the entire field behaves as if success or failure in that mouse determines whether a drug deserves to exist.
This creates a filtering system that is almost perfectly backward.
A treatment that genuinely improves the energy balance of stressed motor neurons in sporadic ALS may show little or no effect in a rapidly degenerating SOD1 mouse engineered to massively overexpress a mutant antioxidant enzyme. Such a compound is often discarded early because it “failed the model.” In practice, the model may simply have been biologically irrelevant to the targeted mechanism.
At the same time, compounds that suppress some narrow downstream feature of the SOD1 mouse phenotype can appear spectacularly successful in preclinical studies even if they have little relevance to the broader ALS population. Those drugs advance into expensive human trials and then collapse because the human disease is not the same disease the mouse had.
The screening model, therefore, selects for treatments that fit the model rather than treatments that fit ALS itself.
This is made worse by the way the SOD1 mouse is constructed. The animals often harbor extremely high copy numbers of mutant SOD1 and develop disease at an accelerated rate. The resulting pathology is compressed, exaggerated, and heavily biased toward oxidative stress and protein toxicity pathways associated with that specific mutation. Human sporadic ALS is usually slower, more heterogeneous, and dominated by TDP-43 pathology instead.
From an engineering perspective, this approach is nonsensical. The field has effectively optimized itself around the wrong transfer function.
A candidate drug that stabilizes mitochondrial energy production, reduces stress granule persistence, improves autophagic efficiency, or decreases neuronal energy expenditure may be exactly what a slowly progressing TDP-43 patient needs. But if that effect does not rescue an overdriven SOD1 mouse racing toward paralysis in a few months, the compound is often terminated before meaningful human data is ever collected.
The consequence is deeply ironic:
- potentially useful drugs are killed because they do not work in a biologically distorted model
- biologically irrelevant drugs survive because they do work in that distorted model
- human trials then fail, reinforcing the illusion that ALS is uniquely untreatable
The problem may not be that ALS is impossible to treat. The problem may be that the filtering system preferentially removes the right answers before they ever reach patients.
***
ALS diagnosis is still often treated as a diagnosis of exclusion. In practice, that means months of ruling out everything else while the disease continues destroying motor neurons uninterrupted. From a biological standpoint, this is absurd.
By the time many patients finally receive a formal diagnosis, significant and irreversible neuronal loss has already occurred. Muscles have wasted, denervation has spread, compensatory mechanisms are failing, and the remaining neurons are already operating under extreme stress. Waiting for certainty may feel medically cautious, but biologically, it often means arriving after much of the damage is already done.
This creates a major problem for both clinical care and research.
For patients, delayed diagnosis means delayed interventions. Even supportive measures matter more early than late. Nutritional optimization, ventilation planning, secretion management, cough assist, energy conservation, communication systems, and prevention of repeated hypoxic or infectious stress all work better before the patient is already collapsing physiologically. The system often behaves as if supportive care can wait until “confirmed ALS,” even though the disease itself does not wait.
For drug trials, the problem may be even worse. Many experimental therapies are probably tested far too late in the disease process. A treatment aimed at reducing protein aggregation, oxidative stress, excitotoxicity, transport failure, or inflammation may have little effect once the damage has progressed enough. That does not necessarily mean the therapy failed biologically. It may simply mean the intervention started after the irreversible phase had already progressed too far.
Imagine evaluating firefighting methods only after most of the building has already burned down. That is roughly how many ALS trials operate.
The exclusion-based diagnostic culture also biases research toward late-stage disease markers, as they are easiest to detect reliably. Earlier, subtle metabolic, inflammatory, transport, or electrophysiological abnormalities may persist for years before classical ALS becomes obvious enough to meet diagnostic criteria. But medicine tends to reward certainty over timing.
In rapidly progressive neurodegeneration, that tradeoff may be backward.
A false positive diagnosis is undesirable. But a perfectly certain diagnosis delivered after massive irreversible neuron loss is useless from a treatment point of view.
***
Prevalence is a bad metric for a disease that kills fast.
Prevalence does not measure how often people get the disease. It measures how many living people have it at a given moment. For a rapidly lethal disease, that number is automatically suppressed by death. Patients disappear from the statistic because they die, not because the disease is rare.
That makes prevalence a perverse metric in ALS. The better the care becomes, the more common ALS appears. Ventilation, feeding support, cough assist, better secretion management, and stubborn refusal to die all increase survival time. Increased survival time increases prevalence. Nothing about that means more people are getting ALS. It means fewer are being removed from the count.
So when ALS is described as rare based on prevalence, the statistic is partly measuring the historic failure to keep patients alive.
Incidence is the more honest number. It asks how many new people get ALS per year. That is the relevant metric for cause, risk, research priority, and drug development. On incidence, ALS no longer looks like some vanishingly exotic curiosity. It looks like a steady, brutal production line of new patients, most of whom are then rapidly erased from the prevalence statistics by death.
This matters because prevalence-based thinking makes ALS look smaller than it is. It hides the turnover. It hides the fact that a new cohort enters the machine every year. It also punishes survival: the moment patients live longer, the disease suddenly appears “more common,” as if survival itself had created a problem.
No. The problem was always there.
Prevalence counts the living queue. Incidence counts the rate at which people are thrown into it. For ALS, incidence is the metric that tells the truth.
***
Clinicians often approach genetic screening for ALS primarily from a hereditary perspective.
The discussion usually revolves around questions like:
“Is this familial?”
“Can your children inherit it?”
“Should relatives be tested?”
Those are important questions, of course. But they are not the only reason genotype matters anymore.
Even today, before definitive cures exist, different ALS genotypes already provide clues about which biological pathways are failing inside the cell.
A mutation in SOD1 points to problems involving oxidative stress, protein misfolding, aggregation, and abnormal handling of reactive oxygen species.
C9orf72 suggests disturbances in RNA processing, nucleocytoplasmic transport, stress granule dynamics, autophagy, and toxic repeat-associated products.
TARDBP directly implicates TDP-43 dysfunction: impaired RNA regulation, nuclear depletion, abnormal phase separation, and aggregation.
FUS again shifts emphasis toward RNA metabolism and transport defects, but through somewhat different mechanisms.
These are not merely academic distinctions. They may determine which therapies have any chance of helping.
A drug aimed at oxidative damage may be more relevant in one subtype than another. A therapy targeting RNA toxicity may make little sense for a disease dominated by protein misfolding. Even supportive strategies may differ if one genotype produces especially aggressive metabolic stress or early respiratory involvement.
Yet many clinicians still implicitly treat ALS as one disease with one pathway, merely divided into “genetic” and “non-genetic” cases.
That framework is becoming outdated.
Increasingly, ALS appears more like a final common syndrome produced by multiple upstream failures. Different genetic variants push motor neurons toward collapse via distinct mechanisms, even if the clinical endpoint appears similar.
This also explains why so many “variant-independent” drug trials fail. If patients with fundamentally different molecular diseases are pooled into a single study population, any real signal may be lost in statistical noise.
The future of ALS treatment is unlikely to be one universal miracle drug.
More likely, treatment will gradually fragment into pathway-specific approaches guided by genotype, biomarkers, and cellular pathology. In that sense, genetic testing is no longer only about predicting inheritance.
It is becoming a way to identify which parts of the machinery are actually breaking down.
***
If your business is selling drugs, a cure is economically awkward. A cured patient ceases to be a customer. A drug that slows progression, prolongs life, or modestly improves function creates a continuing revenue stream instead. From a purely financial perspective, chronic management is often more attractive than eradication.
This does not require a conspiracy. It naturally emerges from the system’s structure.
A pharmaceutical company is rewarded for stable long-term income, predictable markets, repeat prescriptions, and treatments that patients may use for years or decades. A cure works differently. Once the patient is cured, the market for that patient disappears. If the cure is highly effective, the entire market may gradually disappear.
For society, however, the calculation is almost the reverse. A cure returns people to a productive life. It reduces disability costs, hospitalizations, long-term care, caregiver burden, and the enormous economic losses created when people are removed from the workforce. Most importantly, it directly reduces human suffering rather than merely slowing its progression.
This tension becomes especially visible in diseases like Amyotrophic Lateral Sclerosis, where many existing drugs provide only modest benefits. From society’s perspective, eliminating the disease entirely would be incomparably more valuable than extending survival by a few months. But scientifically, finding a true cure is harder, riskier, and more uncertain than developing drugs that slightly alter progression.
As a result, the goals overlap only partially. Patients want restoration. Society wants recovery. Companies are structurally rewarded for persistent treatment.
That does not mean pharmaceutical companies are evil. Without them, many important drugs would never exist. But it does mean society should not automatically assume that market incentives naturally optimize toward cures. Often, they optimize toward keeping disease survivable, manageable, and treatable for as long as possible.
That is why public research, universities, non-profit foundations, and increasingly AI-assisted open scientific work are so important. Society has interests that extend beyond recurring revenue.